Refactoring

1. An application performs 100 eventually consistent reads per second on a DynamoDB table. Each item retrieved is 6 KB in size. How many Read Capacity Units (RCUs) must be provisioned to support this workload?

• A.Provision 50 RCUs because eventually consistent reads allow two 8 KB reads per second per unit.
• B.Provision 100 RCUs because each 8 KB read block requires 2 capacity units, halved for eventual consistency.✓ Correct
• C.Provision 200 RCUs because each 6 KB item requires 2 capacity units regardless of the consistency model.
• D.Provision 300 RCUs because the 6 KB size is rounded up to the nearest 3 KB increment for read calculations.

2. A transactional logging system writes 50 records per second to a DynamoDB table. Each log entry is precisely 3 KB in size. How many Write Capacity Units (WCUs) should the engineer provision?

• A.Provision 50 WCUs because standard write operations support items up to 4 KB without requiring additional capacity.
• B.Provision 100 WCUs because DynamoDB rounds item sizes up to the nearest even kilobyte for write capacity evaluations.
• C.Provision 150 WCUs because each 1 KB block of the 3 KB item requires one write capacity unit per second.✓ Correct
• D.Provision 300 WCUs because standard write operations require double capacity to ensure synchronous multi-AZ replication.

3. A developer uses an exact sequential timestamp as the partition key for a DynamoDB table. Despite having adequate provisioned capacity, the application frequently encounters ProvisionedThroughputExceededException errors during writes. What is the most effective solution?

• A.Add a random suffix to the timestamp partition key to distribute write requests evenly across multiple partitions.✓ Correct
• B.Decrease the provisioned write capacity to force DynamoDB to consolidate the underlying storage partitions.
• C.Switch from provisioned to on-demand capacity mode, which completely eliminates the possibility of hot partitions.
• D.Implement strongly consistent reads instead of eventually consistent reads to balance the request load dynamically.

4. A DynamoDB table successfully handles minor traffic spikes lasting two minutes. However, a recent sustained spike lasting 10 minutes caused severe write throttling, even though the overall traffic volume was similar. Why did the longer spike fail?

• A.The table's auto-scaling policies always require at least 15 minutes to trigger, leaving intermediate spikes unprotected.
• B.Burst capacity is only available for read operations, meaning any write spike longer than one minute will automatically fail.
• C.DynamoDB retains up to 5 minutes of unused capacity for bursting, which becomes exhausted during longer sustained spikes.✓ Correct
• D.The partition key structure automatically throttles sustained spikes to protect the underlying hardware from overheating.

5. A developer needs to automatically delete old session data from a DynamoDB table after 24 hours while strictly minimizing operational costs. Which approach satisfies these requirements?

• A.Implement an AWS Lambda function invoked by EventBridge to scan and delete items, consuming standard write capacity.
• B.Configure a DynamoDB Stream to monitor item age and automatically drop old records using a provisioned read unit.
• C.Enable DynamoDB Time to Live (TTL) on an expiration attribute, which deletes expired items without consuming write capacity.✓ Correct
• D.Use a Step Functions state machine with a wait state for each item, deleting the item via SDK calls at the end.